1. Introduction
Yeast infections vary in severity and can range from superficial infections [
1] to sepsis with deadly potential [
2]. One of the most relevant yeast genera due to its high morbidity and mortality is
Candida. The
Candida genus includes at least 15 species associated with human pathologies [
3]. Most species of the genus
Candida are considered normal inhabitants of the skin and mucous membranes. However, under specific circumstances, these microorganisms have the potential to express virulence factors that make them pathogenic or opportunistic, particularly in settings in which the patient suffers a debilitating clinical condition [
4,
5], or due to the use of invasive devices [
6], or under immunosuppression [
7]. The most severe candidiasis have a nosocomial origin and the source of infection can be both endogenous or exogenous [
8].
Candida albicans is the most frequently reported species causing human infection [
9], but other species are also reported:
C. glabrata,
C. parapsilosis,
C. tropicalis, and
C. krusei, among others [
10]. In recent decades, the number of yeast infections seems to have increased significantly worldwide [
11,
12]. Although
Candida albicans still remains responsible for most yeast infections, non-
albicans species appear to be increasing in prevalence [
13,
14]. The greatest relevance of these recent changes in the distribution and epidemiology of severe
Candida infections lies in the intrinsic differences between each of these species in their susceptibility to antifungal therapies [
15]. In many cases, the identification of
Candida species makes it possible to predict their potential susceptibility to antifungal medications.
In hospital laboratories, the routine identification of yeasts isolated from clinical samples is performed by phenotypic methods. In many low-income countries (LICs), microscopic observation of fungal structures in the clinical sample and culture are still considered the gold standard. Where available, the culture is usually followed by biochemical approaches based on chromogenic media for identifying the infecting species. Although these traditional methods are useful, they have some disadvantages, such as the prolonged time it takes to generate results until the identification of the microorganism is complete. Moreover, they have limited sensitivity and the interpretation of the results can be moderately subjective [
16]. In countries with greater availability of financial resources, clinical laboratories have a greater number of techniques that allow rapid identification of yeast species, such as MALDI-TOF [
17,
18] or real-time PCR-based methods [
19].
Whatever the diagnostic method used, it is very important to quickly identify the species of
Candida responsible for an infection, in order to make timely decisions regarding appropriate therapy, to reduce mortality, control outbreaks, and carry out epidemiological investigations [
20]. There are only two published studies regarding the distribution and frequency of isolated
Candida species from clinical samples (urine and blood) from Honduras [
21,
22]. Consequently, epidemiological information on circulating
Candida species in the country is scarce and insufficient. For this reason, the aim of this study was to detect and differentiate the
Candida species in patients with suspected fungal infection in a third-level hospital in Honduras and compare the concordance between the traditional diagnostic phenotypic techniques used in Honduras with a molecular method based on PCR-RFLP (Restriction Fragment Length Polymorphisms).
2. Materials and Methods
2.1. Clinical Samples and Yeast Isolation Conditions
A total of 167 axenic fungal cultures were prospectively obtained from the clinical laboratory of a tertiary-level hospital in Tegucigalpa, Honduras (Honduran Social Security Institute, IHSS) from January to August 2019. All positive cultures that showed a predominant presence of yeasts were included in the study. Yeasts were cultured from clinical samples that included urine (n = 63), sputum (n = 45), vaginal swabs (n = 18), blood (n = 12), catheters (n = 9), stool/rectal swabs (n = 5), cutaneous secretion (n = 2), otic secretion (n = 2), oral swabs (n = 4), cerebrospinal fluid (CSF) (n = 2), and abscesses (n = 5). These positive cultures were obtained from inoculating the clinical samples in blood agar or in chocolate agar plates, with the exception of blood samples that were cultured in specific culture bottles. The cultures were incubated at 37 °C for 24 to 48 h. The growth of microorganisms in blood cultures was monitored by the automated system BD BACTECTM FX (Becton, Dickinson and Company, NJ, USA) for 5 days. A subset of 46 biological samples used for yeasts isolation was separated for subsequent molecular analysis in order to detect and identify Candida species directly from the clinical samples: sputum (n = 13), vaginal swabs (n = 12), blood (n = 8), cutaneous secretion of the foot (n = 2), otic secretion (n = 2), oral swabs (n = 5), CSF (n = 2), and rectal swab (n = 2). Those samples were refrigerated until further processing. Some clinical samples, including all urine samples, were not used to directly detect yeast DNA due to the hospital’s working algorithm.
2.2. Phenotypic Identification of Yeast Species
Yeast species were phenotypically identified by two techniques. The first method was by culture of isolated colonies in HardyCHROM® (CRITERION®, Hardy Diagnostics, Santa Maria, CA, USA). These media were incubated at 37 °C for 48 h and evaluated based on the color of the colonies according to the manufacturer’s instructions. A dark metallic green colony was interpreted as C. albicans; medium blue to dark metallic blue colonies, with a blue halo, were defined as C. tropicalis; pink to medium pink colonies were C. krusei; medium-sized, smooth, pink-colored colonies, often with a darker mauve center, were presumptively identified as C. glabrata; dry and dark purple colonies were assigned to C. parapsilosis, while the rest of species produced generally small, white- to pink-colored colonies. The second phenotypic method was the Rapid Yeast ID Panel in a MicroScan autoSCAN4® (Siemens Healthcare, West Sacramento, CA, USA). One colony of the yeasts was taken from the HardyCHROM® plate and the concentration of unit forming colonies was standardized using a 0.5 McFarland standard. This identification system is composed of microwells containing several identification substrates. This method yields results in 4 h and is able to identify 42 species of yeasts and 19 species of Candida.
2.3. In Silico Analysis of Restriction Fragments
Amplification sizes and restriction patterns of
Candida species were calculated using the Geneious
® 9.1.7 software (Biomatters Ltd, Auckland, New Zealand). Sequences downloaded from the National Center for Biotechnology Information (NCBI) were trimmed to include target sequences for primers ITS1 and ITS4. The enzyme MspI was used for in silico digestion (
Table 1).
2.4. DNA Extraction
Cells of each culture were lysed with 1000 µL of a buffer composed of 10 mM Tris (pH 8), 1 mM Ethylenediaminetetraacetic acid (EDTA) (pH8), and 100 mM NaCl. This suspension was incubated in a water bath at 100 °C for 2 min and then stirred for 1.5 min at maximum velocity in a micro-mini BeadBeater® system (Bio Spec products Inc., Bartlesville, OK, USA) with 0.5-mm glass beads. Supernatant was transferred to a 1.5-mL vial. One volume of phenol:chloroform (1:1) was added and mixed vigorously. After centrifugation at 13,000 rpm for 10 min, the aqueous phase was recovered and transferred to a new vial. Precipitation of nucleic acids was carried out by adding 1/10 volume of sodium acetate (3M, pH 5.2) and one volume of cold isopropanol. To facilitate the precipitation, samples were centrifugated at 13,000 rpm for 3 min. After careful removal of supernatant, the nucleic acids were washed three times with 300 µL of ethanol 70%. The dried pellets were suspended in nuclease-free water and stored at –20 °C until further use. The concentration of nucleic acids was calculated using a NanoDrop® spectrophotometer (Thermo Fisher Scientific Inc., Waltham, MA, USA).
The DNA from the clinical samples was extracted following a protocol similar to that described above for yeast colonies with some modifications. These modifications included two initial washing steps of the sample in a buffer solution (Tris-HCI 10 mM, EDTA 0.1 M pH: 8) and 4 µL of proteinase k (20 mg/mL) during the lysis step, followed by incubation at 65 °C for 1 h.
2.5. PCR-RFLP
Molecular identification of
Candida species was performed using a widely used method based on PCR-RFLP [
21,
23,
24]. The amplification reaction was directed to the ribosomal region comprising the internal transcribed spacers ITS1 and ITS2 and the
5.8S gene. Amplification conditions were carried out in a volume of 50 µL and included 25 µL of 2X PCR Master Mix (Promega Corp. Madison, WI, USA), 1 µL of each primer at a concentration of 10 µM, and 1 µL of DNA (40 ng/µL) as the template. The sequences of the two universal primers were: ITS1- 5´-TCC GTA GGT GAA CCT GCG G-3´, and ITS4- 5`-TCC TCC GCT TAT TGA TAT GC-3´. A Veriti™ 96-well Thermal Cycler (Thermo Fisher Scientific, Waltham, MA, USA) was used to amplify the DNA according to the following program: 95 °C for 5 min, 37 cycles of 95 °C for 30 s, 56 °C for 30 s, and 72 °C for 30 s, and a final extension at 72 °C for 5 min. Amplicons were visualized in 1.5% agarose gel electrophoresis with ethidium bromide.
After confirming the amplification of the ribosomal region, 10 µL of the product were digested with the MspI enzyme at 37 °C for 2 h with 2 µL of buffer, 0.2 µL of 10 µg/µL acetylated Bovine serum albumin (BSA), and 0.5 µL of the restriction enzyme (10 U/µL) (Promega Corp., Madison, WI, USA). The digested fragments were analyzed on 2% agarose gel and recorded in a BioDoc-It Imaging System (UVP, LLC; Upland, CA, USA).
In order to ensure the integrity of DNA and an absence of the inhibitors from the clinical samples, a region of the human beta-globin gene was amplified as previously described [
25] using the primers PCO3: 5´-ACA CAA CTG TGT TTC ACT AGC-3´ and PCO5: 5´-GAA ACC CAA GAG TCT TCT CT-3´.
2.6. Data Analysis
The Cohen´s kappa (k) coefficient, standard error (SE), and a 95% confidence interval were calculated to compare the agreement between the three methods (MicroScan®, HardyCHROM®, and PCR-RFLP). The molecular method was considered as the standard. In addition, the ability to detect Candida species directly from clinical samples was compared against the result of the culture.
2.7. Long-Term Preservation of Yeast Cultures
All strains were preserved under freezing at −20 °C in YPD medium (yeast extract, peptone, dextrose) and 99% sterile glycerol.
4. Discussion
In the present study, we used three different methods (two phenotypic and one molecular) for the identification of Candida species isolated from 11 types of clinical samples from a tertiary hospital in Honduras. Yeasts isolated by HardyCHROM® were further identified through PCR-RFLP. According to the molecular results, C. albicans complex was the most common species, followed by C. tropicalis, C. glabrata complex, C. parapsilosis, C. krusei, C. kefyr, C. guillermondii, and C. haemulonii complex. Although C. albicans was the predominant individual species, the rest of the non-albicans species contributed 57% of the total.
Several studies reported similar findings when analyzing clinical samples of diverse origins. A study conducted with clinical samples obtained from Iranian hospitals showed a higher frequency of
C. albicans, followed by
C. parapsilosis,
C. glabrata, and
C. rugosa identified by PCR-RFLP [
26]. A second study analyzed five different types of clinical samples in Iran, and showed that
C. albicans was the predominant species, followed by
C. glabrata and
C. tropicalis [
27]. Another study conducted with 11 types of clinical samples from hospitalized patients with suspected fungal infection in Mexico City showed a higher prevalence of
C. albicans, followed by
C. tropicalis and
C. glabrata [
28]. In India, the identification by PCR-RFLP of 150 strains from different clinical samples also showed a majority of
Candida albicans, followed by
C. glabrata,
C. tropicalis, and
C. parapsilosis. Kaur et al. analyzed
Candida strains isolated from six types of clinical samples, and also reported an important predominance of
C. albicans [
29]. Other studies have analyzed a single type of clinical sample, such as blood [
30,
31,
32,
33], urine [
21,
24], vulvovaginal secretion [
34], bronchoalveolar fluid [
35], and gastroesophageal tract [
36]; however, in all of them the predominant species was
C. albicans. Although there are many epidemiological studies that show a trend towards an increase in infections caused by non-albicans
Candida species [
37], it seems that the
C. albicans complex remains the most frequent etiological agent among fungal human infections when considered individually. However, when non-albicans species are considered as a group, these are the majority with respect to
C. albicans.According to our results, the species most commonly isolated from blood was not
C. albicans but
C. parapsilosis (n = 5; 41.7%). This result is interesting because
C. parapsilosis complex species have emerged as an increasing cause of fungemia [
38], and due to its capacity as a skin colonizer that facilitates transmission from health personnel to patients during the manipulation of intravascular catheters [
39]. In urine, however, the most frequent species were
C. albicans complex,
C. tropicalis, and
C. glabrata complex. A previous study in search of the primary cause of genitourinary infection in a hospital in Honduras analyzed 73 yeasts, with the results consistent with the current study [
21].
The second objective of this study was to compare the capacity of two biochemical approaches and a molecular method to identify
Candida species. Most species identifications among the three methods coincided and showed good kappa agreement, especially for the most common species. Many studies have also shown high agreement between molecular findings and traditional phenotypic tests, such as chromogenic media and automated biochemical approaches [
24,
29,
31,
34,
35,
40,
41]. Consequently, the gold standard for the diagnosis of
Candida infections based on biochemical methods seems to be specific enough for a routine hospital diagnosis.
However, when considering PCR-RFLP as a reference standard in this study, we found that HardyCHROM
® and MicroScan
® failed to correctly identify 9 (5.08%) and 15 (8.47%) isolates, respectively. The three species misidentified by these methods were
C. tropicalis,
C. glabrata complex, and
C. haemulonii complex. An important drawback of the phenotypic methods evaluated in this study is its inability to correctly identify species of the
C. haemulonii complex, to which
C. auris belongs.
C. auris is an important emerging pathogen with a high mortality rate in hospitals [
42].
There are several other reports of inconsistency between molecular and phenotypic assays. Zhai et al. [
34] compared the results of a molecular method based on real-time PCR with CHROMagar Candida
®, and detected that 9.3% of inconsistencies were mostly attributable to the phenotypic method. A study in which several methods were compared, including a chromogenic medium and PCR-RFLP, determined a 2.5% inconsistency between the assays [
31]. Jafari et al. compared CHROMagar Candida
® and two molecular assays. According to their results, the concordances (k coefficient) between the phenotypic assay and the molecular methods were 0.87 and 0.89 [
26].
The causes of inconsistencies between methods could be attributed to a limited ability of phenotypic approaches, such as HardyCHROM
®, to identify rare species [
26], nor can PCR-RFLP identify all potential restriction patterns. Another reason could be the presence of more than one species of yeast in a seemingly pure culture. These mixed cultures have been described frequently [
21,
31,
43,
44], and in cases in which one of the species is underrepresented, they may cause confusion in the interpretation of biochemical results. On the other hand, the use of two phenotypic methods can be considered a good practice that provides greater specificity and sensitivity in the identification of species, mainly when there are two or more species co-infecting a patient.
Currently, molecular techniques based on PCR are not recommended nor have they been approved for clinical diagnosis purposes [
45]. Despite this, the third objective of this study was to identify the yeast species by directly extracting the DNA from eight types of clinical samples. Indeed, it was possible to amplify
Candida DNA in all of the 46 samples analyzed. The species identified in the clinical samples mostly coincided with the results of the molecular identification from the cultures (73.91%). In 12 cases, the results were discordant, especially in samples taken from mucous membranes. Since the mucous membranes are normally colonized by commensal yeast species [
46] it is possible that in some cases, the species responsible for the infection shares an ecological niche with other species and this mixture could be responsible for the discrepancies in the diagnosis. In the sterile clinical samples, the results were more satisfactory. In the eight blood samples analyzed, seven revealed the same species, and only one sample identified
C. tropicalis while the culture identified
C. guillermondii. In the two CSF samples, the results were concordant. This last result is interesting since an early, sensitive, and specific diagnosis of the etiologic agent in a sterile sample, independent of the culture, could positively influence the patient’s outcome and survival.